Data Processing

INTERFEROMETRIC SWATH BATHYMETRY and Backscatter

Navigation and Motion

Position data recorded by the Coda-Octopus F190R IMU system were corrected in real-time with the OmniSTAR HP differential navigation service. The IMU also applied real-time motion corrections for heave, roll, and pitch to the vertical component of each position fix. The corrected positions were integrated with the observed bathymetric values to calculate a final position and ellipsoid height representing the elevation of the seafloor with respect to the geodetic reference frame ITRF2005 (ITRF05) across the swath range. At various times during the cruise, difficulty in acquiring and maintaining satellite signal lock was encountered; therefore compromising the Differential Global Positioning System (DGPS) signal. This problem has since been deduced as poor antenna location causing multipathing and possibly a reduced view to the sky. The quality of the vertical accuracy of the ellipsoid was questioned and the first round of data processing results proved to be poor. To resolve the questionable DGPS accuracy for the vertical component a NOAA Tide Zone model was requested for the area (fig. 4). Using this zone model and the downloaded six minute verified tide data from the Pascagoula NOAA Lab tide station (#8741533) and the Bay Waveland Yacht Club tide station (#8747437), the vertical (z) component was referenced to the Mean Low Lower Water (MLLW) datum instead of ellipsoid height.

Soundings and Backscatter

The Systems Engineering and Assessment Ltd., program SWATHplus serves as acquisition software and initial processing software. Preliminary roll calibration data were collected and processed using SWATHplus and Grid Processor software versions 3.07.17. Instrument offset and calibration values were then input into the session file (.sxs). The raw data files (.sxr) were processed using the updated system configuration file that contained the roll calibration values, measured equipment offsets, acquisition parameters, navigation parameters, SOS at sonar head, and the SVP cast data. Any acoustic filtering applied in SWATHplus was also written to the processed data file (.sxp). The initial datum for the swath and backscatter data was ITRF05, which is the real-time acquisition datum of OmniSTAR HP position and navigation data.

All processed data files (.sxp) were imported into CARIS HIPS and SIPS version 8.1 and finalized in version 8.1.7. The tide zone model and associated verified six minute tide data were applied using the CARIS Load Tide tool. The total propagated uncertainty was calculated for each data file. The original sounding data were edited for outliers using the CARIS Swath Editor tool and associated depth filters. Remaining outliers were then deleted manually. A CARIS BASE (Bathymetry with Associated Statistical Error) surface with the associated CUBE (Combined Uncertainty and Bathymetry Estimator) sample surface was created from the edited soundings dataset. A BASE hypothesis is the estimated value of a grid node representing all the soundings within a chosen resolution or grid-cell size (for example, 2 m) weighted by uncertainty and proximity, giving the final value as a "sample" of the data within the specific grid cell. This algorithm allows multiple grid-node hypotheses to be verified or overridden by the user while maximizing processing efficiency.

A 2-m resolution CUBE surface was created to perform initial CUBE hypothesis editing followed by a higher 1-m resolution BASE surface cleaning using the CARIS Subset Editor tool. The 2-m resolution BASE surface sample x, y, z data were exported as ASCII in ITRF05 for the horizontal datum and MLLW for the vertical datum. The data were then transformed horizontally into WGS84 (G1150) and NAD83 referenced to GEOID12A with NOAA's VDatum software conversion tool version 3.2.

The interferometric backscatter (IFB) was processed using SXPTools (build 175). SXPTools is US Geological Survey software that uses command line programs to enhance the quality of the backscatter from SEA SWATHplus processed data files using an empirical gain normalization scheme. The processed (.sxp) files were gridded using MBSystem (Version 5.3) and exported in ESRI ASCII Grid format. The grid was imported into ESRI's ArcGIS Version 10.2, converted to a raster file with the ASCII to Raster Tool. The final mosaic was exported in GeoTIFF format referenced to ITRF05 (fig. 5).

Figure 5. A GeoTIFF image of the interferometric backscatter data. This is a 1-meter grid resolution with an amplitude intensity ranges from -8 to 64. [Click to enlarge].

Sidescan Sonar

The sidescan sonar acoustic backscatter data from the Klein 3900 was imported into SonarWiz5 version 5.06, and bottom tracking was applied creating a slant range corrected record to define the seafloor directly below the swath transducers. Once the seafloor was defined, slant range was applied to remove the water column from the data and minimize target shadows. A backscatter mosaic was created from the individual SSS lines and adjusted for contrast and brightness. The final mosaic was exported in GeoTIFF format and imported into Esri's ArcGIS version 10.2, where the histogram was adjusted to enhance the display of seafloor surface material. Missing tracklines from the mosaic grid are a direct result of equipment issues with the Klein 3900 during the cruise. Data were corrupt, noisy, or not collected in these areas (fig. 6).

Figure 6. A GeoTIFF image of the sidescan sonar data from the Klein 3900. This is a 1-meter grid resolution with an amplitude intensity ranges from 0 to 64. [Click to enlarge].

DIGITAL ELEVATION MODEL

Using Esri ArcGIS, the swath soundings were imported into ESRI’s ArcMap version 10.1 and gridded using the Geostatistical Analyst Tool's "radial basis function". This gridding allowed the user to adjust interpolation parameters with regard to real data spatial resolution and orientation when predicting values. This method produced a less biased representation of the dataset as a whole. A cross-validation report can be generated when using the radial basis function that helps the user understand how well the model will predict data values at locations without data points. In general, the cross-validation takes one point and predicts its position using the weighted surrounding data values. The predicted value is compared against the actual value of that data point and general statistics are computed. The validation report for the swath bathymetry 50-m grid is listed below. The radial basis function surface was exported to a raster file.   Measured Predicted Error Count 11,761,695 11,761,695 11,761,695 Minimum 3.68 3.50 -2.84 Maximum 19.25 20.15 +2.59         A bounding polygon representing the extent of survey tracklines was created and converted into a raster mask using the ArcGIS "polygon to raster" conversion tool. The DEM was then clipped to the raster mask using the ArcGIS Spatial Analyst "extract by raster mask" tool (fig. 7). It is common, in large-area, shallow water surveys, to find wide data gaps between tracklines relative to the swath width surveyed (fig. 2). To minimize data gap concerns in the final DEM, the ArcGIS Spatial Analyst "neighborhood" low-pass raster- data filter was applied. The DEM was converted to ASCII file using ArcGIS "raster to ASCII" conversion tool.

Figure 7.  50-meter digital elevation model for the 2013 Coastal Change and Transport geophysical survey off shore of Petit Bois Island, Gulf Islands National Seashore, Mississippi. [Click to enlarge.] Abbreviation: m, meter.

Note: Any use of trade names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

(back to top)